Multi-step Entity-centric Information Retrieval for Multi-Hop Question Answering

Multi-hop question answering (QA) requires an information retrieval (IR) system that can find \emph{multiple} supporting evidence needed to answer the question, making the retrieval process very challenging. This paper introduces an IR technique that uses information of entities present in the initi...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2019-09
Main Authors: Godbole, Ameya, Kavarthapu, Dilip, Das, Rajarshi, Gong, Zhiyu, Singhal, Abhishek, Zamani, Hamed, Yu, Mo, Gao, Tian, Guo, Xiaoxiao, Manzil Zaheer, McCallum, Andrew
Format: Article
Language:English
Subjects:
Information retrieval
Questions
Retrieval performance measures
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Multi-hop question answering (QA) requires an information retrieval (IR) system that can find \emph{multiple} supporting evidence needed to answer the question, making the retrieval process very challenging. This paper introduces an IR technique that uses information of entities present in the initially retrieved evidence to learn to `\emph{hop}' to other relevant evidence. In a setting, with more than \textbf{5 million} Wikipedia paragraphs, our approach leads to significant boost in retrieval performance. The retrieved evidence also increased the performance of an existing QA model (without any training) on the \hotpot benchmark by \textbf{10.59} F1.
ISSN:2331-8422